Ki67 biomarker distinguished luminal A and luminal B breast cancer

Ki67, a nuclear marker of cell proliferation, could consistently distinguish luminal A breast cancer from luminal B disease.

To determine this, researchers subjected a group of tumors from patients with invasive cancer to gene expression profiling. Another group of hormone-receptor positive tumors was assessed using immunohistochemistry. The immunohistochemistry tested for four biomarkers: ER, PR, HER2, Ki67.

By linking available immunohistochemical data with the expression profile assignments, they identified 84 hormone receptor–positive, HER2–negative tumors as luminal A and 60 as luminal B.

Using receiver operating characteristics curves and a gene expression profile–defined gold standard, the researchers determined that among the 144 tumors, the best cutoff value for Ki67 to determine luminal B disease was 13.25%. Sensitivity for the marker was 72% (95% CI, 59%-82%) and specificity was 77% (95% CI, 67%-85%).

When the researchers restricted the ROC analysis to the 74 unambiguously luminal A and 53 luminal B tumors, the best cutoff value for Ki67 remained 13.25%. Sensitivity improved to 77% (95% CI, 64%-87%) and specificity was 78% (95% CI, 68%-87%).

The researchers selected a Ki67 index of ≥14% Ki67–positive tumor nuclei as the threshold for human visual assessment. They said that defining a visually assessable cut point allowed assignment of standard pathology breast cancer specimens into luminal A and B subtypes using immunohistochemical approaches.

The luminal A subtype was then defined as ER– and/or PR–positive, HER2–negative and Ki67 index <14%. Luminal B subtype was defined as ER– and/or PR–positive, HER2–negative and Ki67 index ≥14%.

“Although we consider breast cancer molecular subtyping by gene expression profiling to be the gold standard, we nevertheless believe that there is an immediate need for a well-defined and validated immunopanels for worldwide clinical diagnostic use,” the researchers wrote.

Cheang MC. J Natl Cancer Inst. 2009;101:736-750.

Typically to divide luminal A and luminal B tumors on a molecular basis, you have to do a microarray analysis, and that can be expensive, time-consuming and impractical on a large-scale basis. What the researchers did here was look at an additional marker, Ki67, to see if that was able to consistently subdivide luminal A tumors from luminal B in tumors that were ER– or PR–positive. They were able to essentially divide their luminal A and luminal B patients based on Ki67 and HER2 status and correlate that with prognostic information, but it was also something they were able to correlate back to the actual molecular analysis on microarray. Frequently microarray analysis is costly, it is time-consuming, etc. The Ki67 is something that is reasonably easy for pathologists to do clinically, despite the fact that there can be a lot of interobserver variability. So to that extent, it is a potentially easier way to help subdivide these tumors. The true question that remains unknown is the clinical application of these results and the ability to base therapy recommendations from this Ki67 data.

– Shannon L. Puhalla, MD

Assistant Professor of Medicine, University of Pittsburgh